tompark

> “A lot of people want local inference,” Altman said. “And somehow, they assume that the problem is solved now. Because we are going to have open-source models. We’re going to be able to run them on our own computers. And so we’re not going to need all the stuff [DRAM] in the data center.”

At first TFA seemed like typical tech news but reading this quote suddenly struck me as satire, as if it came from The Onion.

What might help is if you had some external entity to be dependent upon you getting work done in order for it to get its work done. A job provides this kind of infrastructure to you.

This suggestion only applies if you're a programmer, and assumes you were doing AI-assisted coding before you got laid off... You know how the agentic AI people say you have to get your slow human self out of the ralph loop? Well, put yourself back in the loop.

Whenever your agent harness is blocked, waiting for your input, it should periodically wake and nag you to engage with it. Once you've got this, now something is depending on you to get work done.

If you need an idea for a project to work on, how about a product/service to help people in your exact situation.

Yeah. I hate commutes like that. I had a job where the commute was 30 min bike or bus to Caltrain then 1-1.5 hour train. It was expensive and ate 3-4 hrs/day.

But after being unemployed almost two years, would do that commute again if I had to.

You're right. It's a belief, could be wrong.

A belief: Just as human exceptionalism misled us to believe in a geocentric universe, we'll discover we've been wrong about consciousness too. That is, non-human animals have degrees of consciousness, and we'll see that it's possible for machines too.

A correlated belief: AGI is coming, eventually, AI technology will get there before climate change and economic pressures stop it from happening.

An obvious observation: Large scale upheaval doesn't affect everyone to the same degree or even in the same direction. Some people and companies actually prospered during the Great Depression. People died and suffered during wars and plagues, while others gained advantage and opportunity -- often due to circumstances and chance, but sometimes beliefs/decisions/actions/preparation made a difference.

An example of my bias: during Covid, people told me that remote work was the new normal and moved away. I said in 5 years we'll all be back in the office. I was wrong, but so were they.

The photo credit says iStockphoto but the image looks absurdly fake (transparent cable with glowing lights inside). The istockphoto home page says their library has no AI images but that same site also offers AI-image generation.

I know that commenting on meta aspects of an article is discouraged here but that image is just ridiculous.

One thing about that show is it explored various aspects of the problem, not just blaming student behavior but also parents, teachers and administrators. It illustrated how the problem is quite complex and not just solved by beating up students.

Absolutely. I've had the identical experience hiring artists for game art/UI. Every once in a while you find an artist who is stellar but most everyone else is not that good.

The TFA is pointing to a prompting skills/creativity issue, but I get equally frustrated with hired artists as with genAI image models. Even with skilled human artists (remote/offshore), I find they're best at copying a particular style so I end up having to google for images of the look that we're targeting, then spend hours painstakingly describing the changes needed. Then iterating many times, and with artists in remote time zones that means each iteration takes a whole day. The process is ironically very similar to using AI image models but can take much longer due to time zones.

I immediately looked up whether it had "piss shiver" and it did. It was disappointing when Cecil (Straight Dope) refused to investigate it.

(Don't suppose you'd include orgasmic quiver?)

Synesthesia is multifaceted. When I was a kid I used to feel a shape in my hand whenever I felt a certain emotion. That went away in my 20's. Later as an adult I guessed it was a kind of synesthesia.

Yup. I'm more paranoid about AI training on my chats than most ppl are. While I use frontier models to do research and to build tools, I've tried to avoid discussing details about my projects (e.g. story worldbuilding) except when using local models like Qwen-27B. Of course this slows down progress and has a real opportunity cost, but it would kill all my motivation if I got "scooped" like that.

Is "one of its most prominent users" a typical LLM phrase? Doesn't seem like it to me.